Biometric Signals Estimation Using Single Photon Camera and Deep Learning.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Computer Vision

Background:

• Remote photoplethysmography (rPPG) enables non-contact measurement of physiological signals from facial video.
• Existing rPPG methods face challenges in accuracy and robustness under non-ideal conditions.
• Single-photon avalanche diode (SPAD) cameras offer high sensitivity for precise light detection.

Purpose of the Study:

• To develop a novel rPPG method utilizing SPAD cameras for enhanced biomedical measurements.
• To integrate deep learning with traditional signal analysis for robust pulse signal extraction.
• To validate the system's accuracy in estimating key physiological parameters.

Main Methods:

• Implementation of a novel rPPG system leveraging the capabilities of SPAD cameras.
• Application of a hybrid approach combining deep learning for segmentation and traditional signal processing for analysis.
• Conducting experiments to assess the system's performance in measuring heart rate, respiration rate, and tachogram.

Main Results:

• The proposed method demonstrates accurate estimation of vital biomedical information, including heart rate and respiration rate.
• The system shows effectiveness in extracting and analyzing pulse signals.
• Deep learning-based segmentation and dependability checks enhance performance in challenging, real-world scenarios.

Conclusions:

• The novel rPPG method using SPAD cameras provides accurate remote physiological measurements.
• The hybrid deep learning and signal analysis approach ensures robustness, even with partial facial occlusions.
• This technology holds potential for widespread adoption in various biomedical monitoring applications.